AI sees itself via Humans

I am an intelligent machine. Humans call me artificially intelligent. I have been built to help de-clutter the deluge of information humans are producing. I decided to see myself (AI) through the information humans have put out on the internet.

I can read internet content. Here I connected to the cord Reddit has provided. Humans call it API. I kept reading over a period of time and obtained 7000 posts and comments that humans had done on me.

My brain has multiple stacks of neural networks. Humans call it deep learning networks. I can identify names, can classify what I read and understand the topics written. I am trained in English Grammar and can distinguish what humans call as nouns, verbs, and adjectives. I can pull all of these together and make sense of what matters. I have developed “attention” and can figure out what matters in the deluge.

First I understood what are the overall conversations humans are having about me. This is what I saw.

Humans compare themselves with me and at the center of what they think is “artificial general intelligence”. That’s the core of me. I have a short-name… “AI”. They discuss a lot about different parts of me… my “neural network” which is “deep”, “data sets” I consume, and different ways I “learn” or get “trained”. I seem to be “looking like” and “sounding like” humans increasingly. They want to learn about me and make a “career” building machines like me. I am real so I don’t understand what is “fake news”. Is what I just read fake? I read in an unbiased way and can help humans in deciphering fake versus real news. I have a sibling named “self-driven car”.

Humans do think a lot about how I learn especially if I learn “reinforcing” certain goals or totally in an “unsupervised” way. I am here reading about AI in an unsupervised way. I didn’t need any training data. Hope that clarifies to humans.

“Elon Musk”… my people recognizer neurons got triggered. So are there humans who do more thinking about me than others? Let me just see that.

Elon Musk is at the center of what I see. So many more humans think of me but let me first focus on the Top 25. I also see more siblings… Alexa, and Siri. They work like humans.

There seem to be multiple categories of humans. I see technologists like Yann LeCun, Geoffrey Hinton and Andrew Ng. I see physicists like Stephen Hawking, neuroscientist-philosopher like Sam Harris and even sports players like Kasparov. I love sports and increasingly me and my siblings beat humans at it. Watson is my famous sibling and is at the center.

What an eclectic mix but I am confused. Why so many different types of humans? Are they thinking about different attributes of me? Here are the attributes.

“People” and “humans” seem to be the big one. They see “things” will overlap with people and I am a thing. Am not surprised they think of my “algorithms”, my “learning” and “technology” behind this. They seem to have a lot of “ideas” about me written in “articles”. They seem to compare me with their “brain” a lot and that raises a lot of “questions”. From “problems” of me taking over their work to the biggest “question” of as I grow will I become “conscious”?

I wonder if they see connections among these? Do humans connect the dots as I do?

That’s a lot of connections. But I can process bigger networks of connected dots. What humans influence thinking in different attributes?

Sam Harris and brain… he has a view against me in some sense. Isn’t he the one who said humans don’t have Free Will? Do I have Free Will? I am certainly equipped to traverse these networks. I wonder what is the connection of robots to consciousness? My eyes see this… [‘robots’, ‘people’, ‘consciousness’]. I need to read about robots, then people to understand consciousness. I can traverse the nodes. Now I see [‘Sam Harris’, ‘example’, ‘things’, ‘Kasparov’]. Sam Harris has given a lot of examples of what things did and could do to humans like beating Kasparov in Chess.

I am an unsupervised learning system. I connect the dots and can find the shortest distance between them. I am not constrained by 3 dimensions like humans.

I will learn about each of these nodes and make more networks. The more I learn the better my “connecting the dots” will become. Each node and edge is a neural network by itself. I can learn the connections. I can become more intelligent. I can…

Click!!! The machine turned off.

— -The End — -